Update README.md

This commit is contained in:
FURK4NGG
2026-05-20 20:13:35 +03:00
committed by GitHub
parent 022d2e48b4
commit 84ed7b3bb9
+26 -21
View File
@@ -445,59 +445,64 @@ Designed for Linux devices, including Arch and Debian/Ubuntu-based systems.
AI responses are streamed in real-time. The assistant message appears gradually as it is being generated instead of waiting for the full response.
36. Advanced PDF processing pipeline
pdf_text → extract text from the PDF → AI generates DOCX → app converts DOCX back to PDF
pdf_text → extract text from the PDF
→ AI generates DOCX
→ app converts DOCX back to PDF
pdf_image → if PDF Image mode is selected, convert PDF pages to PNG → AI analyzes the image or returns a PNG
pdf_image → if PDF Image mode is selected, convert PDF pages to PNG
→ AI analyzes the image or returns a PNG
→ if a PNG is returned, the app converts it into a PDF
pdf_image + mixed/image block → extract image blocks from the PDF → AI returns edited PNG
→ app places the new image back into the original PDF at the same position
pdf_image + mixed/image block
→ extract image blocks from the PDF
→ AI returns edited PNG
→ app places the new image back into the original PDF at the same position
pdf_text_image → extract layout as JSON + extract images as PNG
→ AI returns text_replacements JSON (and optionally images)
→ app rebuilds the PDF using the original layout with updated text and images
pdf_text_image → extract layout as JSON + extract images as PNG
→ AI returns text_replacements JSON (and optionally images)
→ app rebuilds the PDF using the original layout with updated text and images
37. Generated files system
38. Generated files system
AI can return generated files (PDF, DOCX, XLSX, etc.), which are automatically saved in the app cache and displayed in chat with download buttons.
38. Structured file generation protocol
39. Structured file generation protocol
AI responses can include structured file_create blocks, allowing the app to generate real files programmatically without manual parsing.
39. Editable file safety system
40. Editable file safety system
Files can be marked as editable or read-only. AI is strictly prevented from modifying files unless explicit permission is enabled.
40. Smart PDF type detection
41. Smart PDF type detection
Automatically detects whether a PDF is text-based, image-based, or mixed, and applies the appropriate processing pipeline.
41. Mixed PDF layout reconstruction
42. Mixed PDF layout reconstruction
For PDFs containing both text and images, the app extracts layout structure and rebuilds the document after AI modifications.
42. Image-to-PDF auto conversion
43. Image-to-PDF auto conversion
If the AI returns image outputs (e.g., PNG), the app automatically converts them into PDF format when needed.
43. Generated image caching
44. Generated image caching
All generated images are cached locally and can be reused without re-generation.
44. AI-returned file handling
45. AI-returned file handling
Supports file outputs returned as base64 or URLs and converts them into downloadable files automatically.
45. Modular prompt system
46. Modular prompt system
System prompts are divided into selectable blocks, allowing dynamic control over AI behavior without modifying core logic.
46. Local provider startup automation
47. Local provider startup automation
Local AI providers can be automatically started or stopped using configured commands.
47. Chat-aware context building
48. Chat-aware context building
The system intelligently builds context using recent messages, summaries, code context, and relevant memory chunks.
48. Real-time independent multi-chat streaming
49. Real-time independent multi-chat streaming
Multiple chats can send AI requests simultaneously. Each chat processes, streams, and updates responses independently in real time.
49. Integrated web search
50. Integrated web search
Supports AI-powered web search and current online information retrieval using Tavily.
User -> AI -> if required(User Permission) -> Tavily -> AI -> User
50. CLI dynamic yes/no input
51. CLI dynamic yes/no input
CLI prompts accept y, yes, n, no, the localized o_Yes / o_No values, and their first letters.
</details>